1 I, pp 2881-2894,
Printed in Great Britain.
1991.
ooo9-2509/91
53.00 + 0.00
CC 1991 Pergaman Press plc
THEORETICAL
AND EXPERIMENTAL
BUBBLE
FORMATION AT A SINGLE ORIFICE IN A
TWO-DIMENSIONAL
GAS-FLUIDIZED
BED
Department
J. A. M. KUIPERS,
W. PRINS and W. P. M. VAN SWAAIJ
Twente University of Technology, PO Box 217, 7500 AE
of Chemical Engineering,
Enschede, The Netherlands
(First received
17 September
7 March 1991)
Abstract-An
earlier developed,
first principles
hydrodynamic
model of gas-fluidized
beds has been
employed to study theoretically
bubble formation
at a single orifice in a two-dimensional
bed. For several
orifice discharge
rates, theoretically
predicted
bubble sizes, formation
times and shapes have been
Besides, the present experimental
compared
with experimental
data obtained from triggered photographs.
and theoretical
results were compared
with predictions
from two approximate
models reported
in the
literature
which are based on an idealized picture of the process of bubble formation.
The advanced
hydrodynamic
model appears
to predict the experimentally
observed diameters, formation times and
shapes of bubbles quite satisfactorily.
The observed and calculated
bubble diameters
fall between the
models. Both the experimental and theoretical results clearly indicate
predictions
from the two approximate
that a strong leakage of bubble gas into the surrounding
porous emulsion phase occurs, especially during
the initial stage of bubble formation.
1. INTRODUCTION
The formation
of gas bubbles is one of the most
characteristic
phenomena
of fluidized beds and it has
been recorded
over a considerable
span of time. Many
unique properties
of fluidized beds can be related
directly to the presence of bubbles and are dominated
by their behaviour. Therefore, accurate prediction
of
bubble characteristics
such as the size distribution,
the
bubble
rise velocity
distribution
and the bubble
frequency distribution
is practically important.
However, these distributions
depend on the initial bubble characteristics
at the gas distributor,
where the
bubbles are generated. Furthermore,
it is now widely
recognized that the gas-solids contacting efficiency of
fluidized bed chemical reactors is quite sensitive to the
bed hydrodynamics
just above the gas distributor
plate. Mass and heat transfer processes are seriously
affected by the mechanism
of bubble formation.
in
large scale gas-fluidized beds, normally, bubbles originate from discrete holes or other orfices in the gas
distributor
plate on which the bed rests. Bubble formation in gas-fluidized
beds at discrete orifices has
been studied experimentally
as well as theoretically
by
a number
of investigators.
Several
approximate
models, based on a strongly idealized picture of the
process of bubble formation,
have been presented in
the literature.
According
to the Davidson
and Schiiler model the
bubble volume Vb at detachment
and the corresponding time for bubble formation
t, (bubble detachment
time) are given respectively by
V, = 1.725(Co)0-h
g
(
>
and
fb
2
Q
(2)
where C, represents
the virtual mass coefficient of a
sphere, 9 the acceleration
of gravity and Q the constant gas flow rate through the orifice. The value of
the virtual mass coefficient C, is geometry dependent
(Davidson
and Harrison,
1963; Milne-Thomson,
lY60) and has been calculated for a number of relatively simple
configurations
of practical
interest
(Fig. 1). In the model adopted by Harrison and Leung
it has been assumed that gas leakage through
the
bubble surface into the emulsion
phase does not
occur. Nguyen and Leung (1972) injected air through
an orifice into an incipiently
fluidized two-dimensional bed of alumina particles. They correlated
the
observed
bubble volumes
with the gas flow rate
through the orifice and the frequency of bubble formation, nb, as
indicating considerable
(47%) leakage of the injected
gas into the emulsion phase during the process of
bubble formation. Rowe et al. (1979) used X-ray cinephotography
to investigate the entry of gas from an
orifice into various fluidized powders. The technique
of X-ray observation
of bubbles in gas-Auidized beds,
J. A. M.
2882
,
orifice
disbibutor
KUIPERSet al.
(b)
orifice
plate
I-
emulsionphase
Fig. 1. Two geometrically different configurations for bubble formation at a single orifice in a gas-fluidized bed (threedimensional case). (a) C, = 1l/16, (b) C, = l/2.
where
Combination
of eqs (4) and (5) yields a differential
equation governing the rate of change of the bubble
diameter D, with respect to time which can be integrated with the initial condition,
D, = 0 at t = 0, to
give the following equation relating the time for bubble formation
t, and the corresponding
bubble diameter D,:
D h.mar=
Here 4, max represents the maximum bubble diameter, at which the total gas leakage through the
bubble boundary equals the total inflow through the
orifice (i.e. the jet). Equations
(6) and (7) and the
experimentally
determined bubble frequency nb = l/r,
provide the relations required to caIculate the bubble
diameter nh and the integral gas leakage 0, defined
by:
R = Qtb -
I(, =
Qt,- ;D:
Satisfactory
agreement
between the model predictions and the experimental
data was found. However,
a major drawback of the model proposed
by Yang
er al. (1984) is the necessary
input of the experimentally
observed
bubble frequency.
This experimental input is required because Yang et al. did not
employ a momentum
balance for the bubble. To
predict theoretically
both the (initial) bubble diameter
and the corresponding
bubble frequency,
the mass
balance
and momentum
balance
of the forming
bubble should be solved simultaneously.
This approach was followed by Caram and Hsu (1986) in a
theoretical model describing the formation of a spherical bubble in a fluidized bed.
Their equation of motion, describing the position of
the bubble centre S as a function of time, was identical
to the one employed in the Davidson
and Schiiler
model. Unlike the Davidson and Schiiler model adopted by Harrison
and Leung for spherical
bubble
formation
in Ruidized beds, the Caram and Hsu
model accounts for gas leakage into the emulsion
phase. Darcys law was used to obtain an expression
for the superlicial gas leakage velocity at the bubble
boundary.
Caram
and Hsu reported
satisfactory
agreement
of their model predictions
with limited
expcrimcntal
data obtained from the litcraturc.
The main objective of the present investigation
is to
compare experimentally
determined
sizes, of bubbles
formed at a single orifice in a two-dimensional
gasfluidized bed, with theoretical predictions
from a previously
developed
first principles
hydrodynamic
model of fluidized beds (Kuipers
et al., 1991). In
addition, the experimental
and theoretical results will
be compared with predictions from two approximate
models, both based on an idealized picture of the
process of bubble formation
in fluidized beds. Some
assumptions
of these approximate
models will be
discussed in relation to the results obtained from the
advanced hydrodynamic
model.
Bubble
formation
&iJ =Q -
Urn,4
(12)
where 6, represents the bed thickness of the twodimensional bed. Equations (11) and (12) can be
solved, by specifying appropriate initial conditions, to
0.57 m
yield the
function
S = R,.
solution
must be
bed
2883
3- flange
l
gas-fluidized
/5izA
I- flange
-gas distributor
section (plexiglass)
- 3 mm glass beads
J. A.
2884
M. KUIPERS et al
(a) Calculation
of the equivalent
bubble
D, on basis of an elliptical bubble shape:
(13)
D, = J%%
Table 1. Hydrodynamic
diameter
4. HYDRODYNAMIC
MODEL
Our previously
developed
theoretical
model of
gas-fluidized
beds (Kuipers et al., 1991) is based on a
two-fluid model approach
in which both phases are
considered
to be continuous
and fully interpenetrating. In fact, the equations employed in this theoretical model can be seen as a generalization
of the
Navier-Stokes
equations
for two interacting
continua.
Two sets of conservation
equations are used, governing the balance of mass, momentum
and thermal
energy in each phase. Table 1 shows the mass and
Continuity equations
Fluid phase
ah4
-
+ (V*Ep,U) = 0
at
(Tl-1)
Solid phase
act1 - E)YSI
+ [v*(l
- E)P,V]
at
(Tl-2)
0.
Momentum equations
Fluid phase
4EPA
at
+ (v*&p,UU)= -
2
Evp
p(U
V) +
i+(v*U)~
II
(U-3)
Solid phase
au1 ~ E)P,VI
at
v (I - E) - &(V.iI
i.[
- G(E)VE+ (I-
II
E)p,g (Tl-4)
Bubble
momentum
conservations
equations for both phases
in vector form. In the present study bubble formation
in a cold-flow two-dimensional
gas-fluidized
bed-will
be studied and because of the anticipated
small heat
effects, the solution of the thermal energy equations is
not considered
here. Due to the mathematical
complexity of the equations of change, a numerical solution method has been used (Kuipers et al., 1991).
The numerical technique has been embodied in an
unsteady two-dimensional
computer code written in
VAX-PASCAL.
The computer
model calculates the
porosity,
the pressure, the fluid phase temperature
and the solid phase temperature
and the velocity fields
of both phases in two-dimensional
Cartesian or (axisymmetrical)
cylindrical coordinates.
These variables
constitute the so called primary or basic variables.
For closure of the set of balance equations specification of the constitutive
relations is required which
implies specification
of all other variables in terms of
the basic variables. Incorporation
of these constitutive equations introduces
the necessary empirical information.
4.1. Constitutive equations
Fluid phase density p/ and solid phase density pS:
The fluid phase density is related to the pressure and
Buid phase temperature
by the ideal gas law:
Mr
pr
=mpFor the solid phase microscopic incompressibility
was
assumed. Accordingly a specified constant density p.,.
was taken:
(16)
P, = P&O.
Interphase
momentum
transfer coefficient 0: For
porosities E -c 0.8 the interphase
momentum
transfer
coefficient has been obtained
from the well-known
vectorial Ergun equation (see Radestock and Jeschar,
1971):
p=
150(l
-7 - 8
E
Pr
(U,)
gas-fluidized bed
where
Re
(184
c
d
U8W
to the Reynolds
;[IP +0.15(Rep)0.687],Re,
< 1030
0.44,
P-
Re,
EPf
lu - vld,
p,
(184
study:
p, = 2.0 x 10e5 Pas.
(19al
Table 2. Apparent
whereas
for porosities
E > 0.8 the
interphase
momentum
transfer coefficient has been derived from
the correlation
of Wen and Yu (1966):
VI
(17)
f(E) = &-2.65.
2885
Vc)
Particle diameter
fuml
550
460
220
170
140
120
82
60
Apparent
Results
of S&tiger1
size range
bed viscosity
Pas
Results
of Grace
Pas
0.95
1.20
0.90
0.80
0.80
0.80
0.60
0.40
0.85
0.75
0.80
0.85
0.90
0.70
$Deduced
from bubble
shape measurements.
J. A. M. KUIPERS et al.
2886
/L~
1.0 Pas.
(19h)
-&)I}
0.402
0.250 m/s
5.0, 10.0, 15.0 m/s
Particle diameter
Particle density
5.00x IOe4m
2660 kg/m3
Orifice diameter
Bed width
Initial bed height
1.50 X to-2m
0.51 I11
0.50 m
101,325.o
X-grid size
Y-grid size
7.50 x lO-m
Time step
2.50 x 10-4s
1.25 x lo-rn
(20)
Pa
initial conditions
E=l
p=pO
ux= 0 y= Umf
vx=v
E=&
mf
=o
x
=o
Cmf
y
5nf
=V=O
7.
Y
P =P,,+
prescribed
ta
continuative
(I-~,~)(p,-p~,,)g(h~i
rigid wallfor
solid phase
of bubble formation
in a cold-flow
Bubble
Fig. 4. Photographically
formation
gas-fluidized
bed
2887
2888
J. A. M.
KUIPERS
et al.
Theoretically
calculated
and experimentally
observed bubble formation
for three different orifice
velocities, namely u, = 5, 10 and 15 m/s will he discussed. In addition, the theoretical
and experimental
results will be compared
with two approximate
models reported
in the literature.
Detailed results
however
are presented
only for the case of U,
= lOm/s.
5.1. Detailed resultsfor u, = 10 m/s
Figure 4 shows a comparison
of photographs
with
corresponding
density
plots for U, = lOm/s.
The
measuring grid visible in the photographs
is the same
as the computational
grid used for the numerical
simulations
(6x = 0.0075 m, 6y = 0.0125 m). The
density plots, shown in Fig. 4, have been obtained
from the calculated
instantaneous
solidity distributions according
to the procedure
described
in the
earlier paper (Kuipers et aE., 1991). In these plots the
local dot density is a measure for the local instantaneous solidity (1 ~ c).
A reasonable
similarity between the photographs
and the density plots can be observed. However, the
agreement between the photographs
and the density
plots is not perfect, especially near the roof of the
bubble where some discrepancies
between experiment
and theory can be observed. The photographs
show
subtle deviations from symmetry about the heart line
of the fluidized bed and aIso show a much sharper
porosity
transition
near the bubble roof. A better
resolution near the roof could probably be obtained
by using a much finer computational
grid; however, to
keep the computer
time reasonable,
this was not
attempted
in the present study. When judging the
theoretical
results depicted in Fig. 4 it must first be
borne in mind that the present model contains no
adjustable
parameters,
and second that, unlike previous theoretical
approaches
(Harrison
and Leung,
1961; Zenz, 1968; Caram and Hsu, 1986) no specific
assumptions
concerning
the mechanism
of bubble
formation have been made. Further inspection of Fig.
4 shows that, especially
during the final stage of
bubble formation,
both the theoretical
and experimental bubbles have a practically circular shape. A
comparison
of the theoretical
and experimental
results for the other orifice velocities (i.e. 5 and 15 m/s)
CM)
I
I
-0.10
-0.05
0.00
0.05
0.10
(21)
To determine the bubble size as a function of time,
during the process of bubble growth, each 10 ms
triggered photographs
of the bed were taken. For the
interpretation
of the experiments
it was found necessary to correct for the time delay (approximately
20 ms) caused by the magnetic valves. As mentioned
earlier two procedures
for the evaluation of the equivalent bubble diameter D, from the photographs
were
considered
[see eqs (13) and (14)]. During the final
stage of bubble formation
the deviation
from the
2889
gas-fIuidized bed
+ Experimental data
Fig. 6. Theoretically calculated and experimentally observed bubble growth at a single orifice in a twodimensional gas-fluidized bed (u, = 10 m/s). The predictions according to two approximate literature
models, which are based on an idealized picture of the bubble formation process, are also shown.
E > 0.85
E > 0.90
0.183 m
0.201 In
0.181 m
0.172 m
0.182 m
0.164 m
0.156 m
0.155 m
0.143 m
E >
D,
D
0,
been neglected.
For the case that the superficial
leakage velocity at the bubble boundary
equals the
superficial minimum fluidization
velocity (second approximate model), the bubble diameter as a function
of time has been obtained from integration
of the gas
mass balance eq. (11) to yield
hnf t
D?
+D,=__
Dbm
for
t < tb
(24)
where
circular bubble shape was always very small (D, x D,,)
and both procedures
yielded essentially
the same
results. However, due to the flat bottom of the bubble
during the initial stage of bubble formation,
the
assumption
of an elliptical bubble shape resulted in a
consequent
underprediction
equi-
valent
for
formation
t, = [ (F)($$)].
t-ctb
(22)
tb is given by
(23)
%4
Db = -.
Equation
(25) defines
the maximum
attainable
bubble diameter achieved when the total gas leakage
through the bubble boundary equals the total inflow
through
the orifice. However,
at the condition
of
bubble detachment
(S = Rb) the bubble diameter is
always smaller than the asymptotic
diameter calculated according
to eq. (25) which implies that the
situation of maximum leakage does not occur. The
time for bubble formation
t, has been obtained
by
simultanteous
numerical
integration
of eqs (11) and
(12) with a fourth order Runge-Kutta
method, S = R,
being the imposed condition
for bubble detachment.
However, it must be noted that the computed bubble
growth curves from the approximate
models, shown
in Fig. 6, have been obtained from the solution of the
respective gas mass balances only. In fact, according
to the approximate
models, bubble growth terminates
at t = tb (i.e. at t = 0.20s and t = 0.15 s for the model
with and without gas leakage, respectively) due to the
bubble detachment
from the orifice and the process of
bubble formation
repeats itself. Inspection
of Fig. 6
shows that the agreement
between the predictions
from the advanced
hydrodynamic
model and the
experimental
data is satisfactory.
The hydrodynamic
model tends to predict somewhat larger bubbles than
the experimentally
observed
ones which may be
caused by the wall effects in the two-dimensional
bed.
These wall effects (due to the presence of the front and
J. A. M.
2890
KUIPERS
as
Y(t) = 1 -
b() _
Yo(Q
for a two-dimensional
the integral
gas leakage
geometry
from a growing
bubble,
(27)
with
el
al.
+ Experimental data
Q = uadz
Fig. 7.
Integral
leakage fraction Y(t), calculated from eq. (27), as a function of time for bubble growth at a
single orifice in a two-dimensional gas-fluidized bed (u, = 10 m/s).
Bubble
o.30 Figure 8a
gas-fluidized
bed
2891
ii = (sin(Q), cos(@))
(4
0.25
t
Y
0.20
0.15
0.10
0.05
0.00
-0.20 -0.15 -0.10 -0.05 -0.00 0.05 0.10
x +
(m)
0.00
0.20
0.40
0.60
0.15
0.80
0.20
1.00
Fig. 8. Calculated
gas phase mass flux profile at the boundary of a circular bubble with centre S and radius rb approxmating the calculated bubble.
Fig. 9. Experimental
the net local mass efflux J(4) of the gas phase at the
bubble boundary depends only on the angle d, (Fig. 8)
and is given by:
J(4)
= (Epfu.n)
= (~~~u,)sing
Table
+ (~~,u,)cos&
(28)
DV
0,
(photograph)
and theoretical
plot) bubbles.
5. Bubble
diameters
(density
at t = 0.2 s
Experimental
Theoretical
0.174 m
0.164 m
0.170 m
0.172 m
0.182 m
0.164 m
2892
+ Experimental
- Numerical
lj++_
,.,.,.,.
0.00
0.10
0.05
0.20
0.15
t
shape factor
cr during
bubble
__)
data
model
(
0.25
+ Experimental
n
0.0
5.0
15.0
10.0
u0 +
Fig. 11. Equivalent
bubble
diameter
Numerical
gas-fluid&d
data
model
20.0
(m/s)
D, as a function
Q = ucd;
of the orifice velocity
U,
Experimental
5
10
15
0.160
0.170
0.180
0.096
0.145
0.188
Numerical
5
10
15
0.150
0.188
0.213
0.089
0.160
0.217
5
10
15
0.158
0.200
0.228
0.123
0.195
0.256
5
10
15
Cl.112
0.150
0.178
0.070
0.124
0.171
Approximate
leakage
Approximate
model with leakage of gas at u,, level
Bubble
formation
6. CONCLUSIONS
J(4)
VEG-Gasinstituut
the numerical
solution
of the approximate
theoretical
mo-
dels.
NOTATION
D,
bubble surface, m2
virtual mass coefficient
drag coefficient
compaction
modulus
bubble diameter, m
eouivalent bubble diameter,
horizontal
D,
A,
CO
Cl
C
D,
Do
L,
bubble
diameter,
M
n
nb
Q
R
Rb
Rep
rb
v,
VCI
V
X
gas-fluid&d
bed
2893
mass flux of the gas phase at the bubble boundary defined in eq. (28), kg/(m* s)
molecular weight, kg/kmol
unit outward normal vector at bubble boundary
bubble frequency, s - 1
pressure, Pa
gas flow rate through
orifice, m3js
gas constant,
J/(kmol K)
bubble radius, m
particle Reynolds number
equivalent bubble radius, m
bubble centre
temperature,
K
time, s
bubble formation time, s
fluid phase velocity, m/s
minimum fluidization velocity, m/s
superficial
injection
velocity through
orifice,
m/s
bubble volume, m3
bubble volume for the case of zero gas leakage,
m3
solid phase velocity, m/s
lateral coordinate,
m
vertical coordinate,
m
Greek
letters
_
volumetric interphase momentum transfer coefcr
ficient, kg/(m s)
bed thickness, m
lateral computational
cell dimension, m
6x
vertical
computational
cell dimension, m
6Y
E
porosity
minimum fluidization porosity
M
e*
compaction
gas phase volume fraction
shear viscosity, kg/(m s)
cc
density, kg/m3
P
bubble shape factor defined in Fig. 10
angle defined in Fig. 8
;
sphericity
6%
integral leakage fraction defined in eq. (27)
r(t)
integral gas leakage during process of bubble
n
formation,
m3
Subscripts
b
bed
bubble
equivalent
e
emulsion phase
fluid phase
f
h
horizontal
max maximum
mf minimum fluidization
microscopic
property
0
particle
P
solid phase
vertical
V
x-direction
X
y-direction
freeboard conditons
0
conditions
2894
Superscripts
T
transpose
*
asymptotic
J. A. M.
KUIPERS
value
Operator
V
gradient
Vdivergence
REFERENCES
Caram, H. S. and Hsu, K. K., 1986, Chem. Engng Sci. 41,
1445.
Davidson,
J. F. and Schiiler, B. 0. G., 1960, Trans. lnstn
them. Engrs 38, 335.
Davidson, J. F. and Harrison, D., 1963, Fluidized Particks.
Cambridge University Press, Cambridge
Grace, J. R., 1970, Can. J. them. Engng 48, 30.
Harrison, D. and Leung, L. S., 1961, Trans. lnstn ckem. Engrs
39, 409.
Kuipers, J. A. M., van Duin, K. J., van B&urn, F. P. H. and
van Swaaij, W. P. M., 1991, A numerical model of gas-
el al.